Are you a frequentist or a Bayesian?

Don’t know don’t care.

But seriously, I have just known about these two after seeing your cartoon, and the two are a bit too complicated for me to understand.

Bayesian. Probably.

The Bayesian certainly! It’s the $50 bet that matters. If you win, you walk away with $50. If you lose, it doesn’t matter. Why not bet a million dollars?

@stanleybmanly Because the comic is funnier with $50.

I am a believer in the 188th rule of Acquisition:
Never bet on a race you haven’t fixed.

ragingloli, I guess that makes you a Ferengian?

In the comic, is that really the logic a frequentist would use? To me it seems that the probability would have to be calculated as probability of the sun not exploding now times probability of rolling double sixes. This is how we calculate the probability someone has a disease when they get a positive test, for example: one minus probability they don’t have the disease times probability of a false positive. That’s why sometimes when you get one positive test it’s still more likely that you’re not sick, because of the rarity of the disease.

Why don’t they just wait eight minutes?

@Seek Because then he’d be out 50 bucks

@Mariah Yeah, you can’t read too much into the comic. It’s after all just a joke. The concept, though, is real and is a problem sometimes in medicine. I see this all the time especially in alt med studies such as homeopathy or acupuncture. The way a particular study is designed there may be a p <.05 and they alt med people will trumpet it as “proof” of their remedy. But if you put it in a Bayesian lens it makes no sense.

Ah okay. Just trying to understand the difference more. I hadn’t heard the terms, so I Googled it, and I’m still not certain I totally understand. It kind of sounds like Bayesian thought just means applying common sense on top of the data, which I think is dangerous. After all, it was once “common sense” that the Earth was the center of the universe.

@Mariah No, it’s not “common sense” per se, but prior probability. Your Bayesian prediction is only as good as your “prior”, and rigorous thought looks at the priors in terms of the science. There is a bit of intuition involved, which is why the frequentists hate it. But it was Bayesian statistical theory that gave Turing the insight to crack Enigma.

In the comic, which is an absurd example (but it’s a comic), the frequentist looked at the data which showed that the sun was going to explode. The data said that it was, therefore the sun is going to explode. The Bayesian looked at the prior probability of it and said, “I’ll bet 50 bucks it’s bullshit”.

Okay, time to sign off. Game of Thrones is about to start. :-)

Cool, thanks for the explanation, that makes lots more sense now :)

Fortunately I do not need to make such decisions. Both methods have strengths and weaknesses. Bayesian probability uses priors. Frequentists seem like they get things backwards. They assume the form of the distribution and then choose the distribution to match the results. It is like a farmer assuming the reason that the pig was able to get out of the pen was because it was able to fly, since that would give a 100% chance of the pig getting out.